Bugzilla – Bug 64493
VUL-0: CVE-2004-1234: recent ELF path error fix also fixes a DoS
Last modified: 2021-10-27 11:37:40 UTC
Date: Wed, 22 Dec 2004 12:08:29 -0200 From: Marcelo Tosatti <marcelo.tosatti@cyclades.com> To: Mark J Cox <mjc@redhat.com> Cc: vendor-sec@lst.de Subject: Re: [vendor-sec] 2.4 load_elf_binary error path flaw On Tue, Dec 21, 2004 at 01:19:44PM +0000, Mark J Cox wrote: > Chris fixed a flaw found by Kirill Korotaev on April 9th 2004. > http://linux.bkbits.net:8080/linux-2.4/cset@4076466d_SqUm4azg4_v3FIG2-X6XQ > Therefore it affects <2.4.26 > > Anyway it got reported to us with a reproducer that can cause a crash, > https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=142965 > > So since we're going to fix it and it's now a proven local DoS I've > assigned it CAN-2004-1234 The recent binfmt_aout v2.6 backport changes also fix a DoS: ChangeSet@1.1527.1.13, 2004-12-16 16:06:31-02:00, chrisw@osdl.org [PATCH] a.out: error check on set_brk It's possible for do_brk() to fail during set_brk() when exec'ing and a.out. This was noted with Florian's a.out binary and overcommit set to 0. Capture this error and terminate properly. ChangeSet@1.1527.1.16, 2004-12-17 21:45:58-02:00, chrisw@osdl.org [PATCH] Backport of 2.6 fix to insert_vm_struct to make it return an error rather than BUG(). Backport of 2.6 fix to insert_vm_struct to make it return an error rather than BUG(). This eliminates a user triggerable BUG() when user created a large vma that overlapped with arg pages during exec (could be triggered with a.out on i386 and x86_64 and elf on ia64). Signed-off-by: Chris Wright <chrisw@osdl.org>
<!-- SBZ_reproduce --> I dont know whether we included this, so I added it here for clarification. Might be that it can be closed soon if we include it. I was concerned about the word *also* in his mail.
Where can I get the fix he is talking about? Those bk numbers do not mean anything to me...
This was the mail as we got it. No link to the patches. One has to ask either him or kernel folks. I thought 'ChangeSet@1.1527.1.13' is something unique within the newest kernel tree.
Created attachment 27381 [details] binfmt_aout_do_brk_fix_24.patch this is from 2.4 bitkeeper: http://linux.bkbits.net:8080/linux-2.4/cset%401.1527.1.13?nav=index.html|src/.|src/fs|related/fs/binfmt_aout.c
Created attachment 27382 [details] binfmt_aout_do_brk_fix.patch this the same fix from 2.6 mainline: http://linux.bkbits.net:8080/linux-2.6/cset%401.2034.36.23?nav=index.html|src/.|src/fs|related/fs/binfmt_aout.c
andrea, some memory management do_brk magic fixes from mainline 2.4/2.6 kernels.... can you have a brief look at how they interact with the do_brk() fixes...
the overlapping VMA problem was fixed already by us I think
a_out fixes have CAN-2004-1074 vm overlap fixes have CAN-2005-0003
I'm puzzled by this report now. Is there any action required?
aout stuff is fixed in 2.4. aout stuff is also fixed in 2.6. so non issue, we got it already.
CVE-2004-1234: CVSS v2 Base Score: 2.1 (AV:L/AC:L/Au:N/C:N/I:N/A:P)